CN100461862C - Video coding system and its method - Google Patents

Abstract

Said invention provides a video message encoding method encoding method and system for encoding video message sequence. Video message sequence containing N subsequences, and each subsequence containing plurality of pictures. When present invention encoding j picture of i subsequence in video message sequence, the pictures before j picture are all completing encoding. Based on said completed coded picture it generates an initial quantisation scale. Proceeding encoding to said j picture using intermediate code mode According to initial quantisation scale. Judging whether there is a variable scene comparing j picture to ( j -1 ) picture, if above-mentioned result is, then to generate a regulative quantisation scale base on initial quantisation scale, and re - encoding j picture using Interior encode model encode model according to adjusted quantisation scale.

Description

Video coding system and method thereof

Technical field

The present invention relates to a kind of video coding system (Video encoding system) and method thereof, particularly relate to a kind of video coding system and method thereof that adaptability once changes Bit-Rate Control Algorithm (Adaptive one-passvariable bit rate control) and scene change-detection (Scene change detection) that have.

Background technology

In recent years, digital compression (Digital compression) technology has been widely used in compressing various video files, to save transmitting bandwidth (Transmissions bandwidth) or memory capacity (Storage size).In general, present compress technique all is adopting international standards tissue (International organization for standardization, ISO) Motion Picture Expert Group of being formulated (Motion picture experts group standard, MPEG standard), for example: MPEG-1, MPEG-2, MPEG-4,,, etc.

In the file technology, standard MPEG video information coder (MPEG compliant video encoder) is to utilize a target bit rate (Target bit rate), and the video signal sequence of an input is compressed.Above-mentioned target bit rate is to set according to the employed communication channel frequency range of a user, and the video signal sequence after the compression promptly is to be that transmit on the basis with the target bit rate, or is stored in the storage device.Yet via the fixing video signal scene that target bit rate compressed, when decoding, according to the complexity (Scene complexity) of scene, its video signal quality will have very big difference.

In practical application, can utilize change bit rate (Variable bit rate) technology that the video signal sequence is compressed.So-called change bit rate is the complexity at scene, promotes the video coding quality with allocation bit rate suitably.Basically, the Bit-Rate Control Algorithm that is used for video coding can be divided into and repeatedly changes bit rate (Multi-pass variable bit rate) and single change bit rate (Singlepass variable bit rate).Repeatedly changing bit rate is to utilize the complexity information of picture in first front transfer (Preceding passes), reaches the restriction of target bit rate.Single change bit rate then is in the process of coding, the bit rate when dynamically adjusting coding.Can provide the preferably quality of image though repeatedly change bit rate, yet also need bigger memory capacity and, therefore repeatedly change the technology of bit rate and do not meet real-time demand than complex calculations.In addition, in the video signal sequence, when scene changed suddenly, single change bit rate can't provide the quality of image of self-consistentency.

Therefore main purpose of the present invention is to provide a kind of video coding system and method thereof that adaptability once changes Bit-Rate Control Algorithm and scene change-detection that have, to address the above problem.

Summary of the invention

A purpose of the present invention is to provide a kind of video coding system and method thereof that adaptability once changes Bit-Rate Control Algorithm and scene change-detection that have, whether in order to the picture that detects in the video signal sequence is a scene change, and according to the video signal complexity information to this picture recompile in addition, make the coding quality of whole video signal sequence reach consistent.

Video encoding method of the present invention (Video encoding method) is in order to encode to a video signal sequence (Video sequence), and the video signal sequence comprises N subsequence, and N is a natural number, and each subsequence all comprises a plurality of pictures (Frame).That is video encoding method of the present invention is in order to in the video signal sequence, and j picture in i subsequence encoded.Picture before this j the picture in this i subsequence, in this video signal sequence has all been finished coding, and i is the integer index in 1 to N scope, and j is between 2 integer indexs to the number of pictures scope of this i subsequence.

Video encoding method of the present invention includes following steps.Based on the described picture of having finished coding, produce an initial quantization scale (Quantization scale).According to the initial quantization scale, j picture in i the subsequence encoded with one first coding mode.Judge whether j picture in i the subsequence (j-1) individual picture in i the subsequence on scene is a scene change (Scene change).If above-mentioned result then produces a quantitative calibration of adjusting based on the initial quantization scale, and according to adjusted quantitative calibration, j picture in i the subsequence is encoded again with one second coding mode for certainly.

In the present invention, when the present picture of the I in the video signal sequence was a scene change, meeting of the present invention was carried out recompile according to the complexity information of before having finished coded picture to this present picture.In other words, the present invention only can carry out the operation of recompile at the picture of occurrence scene variation.Therefore, the present invention combines the technology that repeatedly changes bit rate and single change bit rate, makes video sequence be listed in through behind the coding, can obtain preferable, the consistent quality of image, and not need a large amount of memory spaces to come the memory encoding data.

Can be about the advantages and spirit of the present invention by detailed description of the present invention being further understood below in conjunction with accompanying drawing.

Description of drawings

Fig. 1 is the functional block diagram of video coding system of the present invention.

Fig. 2 is the calcspar of scene change detector shown in Figure 1.

Fig. 3 is the surveyed area schematic diagram that decision module determined shown in Figure 2.

Fig. 4 is the flow chart of video encoding method of the present invention.

The reference numeral explanation

10: video coding system 12: the video signal sequence

14: input 16: encoder

17: scene change detector 18: the bit stream buffer

20: picture bit counter 22: video signal complexity estimator

24: quantitative calibration generator 171: decision module

172: judge module

Embodiment

See also Fig. 1, it is the functional block diagram of video coding system 10 of the present invention.One video signal sequence 12 comprises N subsequence, and N is a natural number, and each subsequence all comprises a plurality of pictures (Frame).In each picture of video signal sequence 12, the row of pre-defined some and row (Predefined number of rows and columns), each row with each the row all comprise a plurality of macro zone blocks (Macroblock, MB).Video signal sequence (Video sequence) 12 inputs to video coding system of the present invention (Video encoding system) 10 via an input (Input end) 14.Video coding system 10 is in order to encoding to j picture in i the subsequence, and the picture before j the picture in i subsequence has all been finished coding.I is the integer index in 1 to N scope, and j is between 2 integer indexs to the number of pictures scope of i subsequence.

As shown in Figure 1, video coding system 10 of the present invention comprises an encoder (Encoder) 16, a scene change detector (Scene change detector) 17, one bit stream buffer (Bitstream buffer) 18, one picture bit counter (Frame bit counter) 20, one video signal complexity estimator (Video complexity estimator) 22 and one quantitative calibration generator (Quantization scale generator) 24.Encoder 16 is to be coupled with input 14, in order to receive and coding video signal sequence 12.Encoder 16 needs decision earlier in order in this present picture of encoding, the coding mode of each macro zone block (Encodingmode) when a present picture of coding video signal sequence 12.Coding mode can be divided into two types haply: in-line coding pattern (Intra mode) and intermediate code pattern (Inter mode).After a picture is via the in-line coding pattern-coding, then be called in-line coding picture (Intra frame, I picture).After a picture is via the intermediate code pattern-coding, then be called intermediate code picture (Inter frame, P picture or B picture).The I picture with reference to any picture, does not get a picture coding.The P picture then with reference to last picture, gets a picture coding.The B picture then is with reference to last picture and back one picture, and a picture coding is got.

Scene change detector 17 is coupled in encoder 16, and in order to judge j picture in i the subsequence on scene, whether (j-1) the individual picture in i subsequence is a scene change (Scene change).Quantitative calibration generator 24 is coupled in encoder 16, and based on the described picture of having finished coding, in order to produce an initial quantization scale (Quantization scale).As if j picture in i the subsequence is a scene change, then quantitative calibration generator 24 can produce a quantitative calibration of adjusting based on the initial quantization scale, and encoder 16 is encoded with one second coding mode to j picture in i the subsequence again according to adjusted quantitative calibration.Second coding mode is above-mentioned in-line coding pattern.Otherwise, not a scene change as if j picture in i the subsequence, encoder 16 can be encoded with one first coding mode to j picture in i the subsequence according to the initial quantization scale.First coding mode is above-mentioned intermediate code pattern.

In one embodiment of the invention, a video signal sequence 12 comprises 3 subsequences, and each subsequence all comprises 100 pictures.Video coding system 10 is encoded to the 33rd picture in the 2nd subsequence, and the picture before the 33rd picture in the 2nd subsequence has all been finished coding (that is the 1st to the 32nd picture in 100 pictures in the 1st subsequence and the 2nd subsequence all finished coding).Be noted that first picture in each subsequence is encoded with the in-line coding pattern.Quantitative calibration generator 24 can be based on the described picture of having finished coding, to produce an initial quantization scale.The 33rd picture in the 2nd subsequence of 17 judgements of scene change detector is on scene, and whether the 32nd picture in the 2nd subsequence is a scene change.As if the 33rd picture in the 2nd subsequence is not a scene change, and encoder 16 can be encoded with the intermediate code pattern to the 33rd picture in the 2nd subsequence according to the initial quantization scale.As if the 33rd picture in the 2nd subsequence is a scene change, then quantitative calibration generator 24 produces a quantitative calibration of adjusting based on the initial quantization scale, and encoder 16 is encoded with the in-line coding pattern to the 33rd picture in the 2nd subsequence again according to adjusted quantitative calibration.In other words, the present invention only can carry out the operation of recompile at the picture of occurrence scene variation.

Bit stream buffer 18 is coupled in encoder 16, the bit stream in order to temporary picture after encoded.Picture bit counter 20 is coupled in encoder 16, in order to continue to follow the trail of and the coding that adds up after the bit length of each picture of video signal sequence 12, to produce the bit stream (Accumulatedbitstream) that adds up.Video signal complexity estimator 22 is coupled between picture bit counter 20 and the quantitative calibration generator 24, in order to the bit stream that adds up of reception picture bit counter 20 generations, and produce one first video signal complexity (Video complexity) and one second video signal complexity.

The first video signal complexity X _ADetermine via following formula one:

Formula one: $X_A=S_A*Q_A*\frac{F}{N_A^2}$

In the formula one, S _ARepresent one first bit length sum total (Summation of bit length), first bit length sum total is the bit length sum total about all pictures before j the picture in i the subsequence; Q _ARepresent one first quantitative calibration sum total (Summation of quantization scale), first quantitative calibration sum total is the quantitative calibration sum total about all pictures before j the picture in i the subsequence; N _ARepresent one first number of pictures, first number of pictures is the number of pictures about all pictures before j the picture in i the subsequence; F represents a predetermined picture rate (Framerate).In the above embodiments, S _A, Q _AAnd N _ABe respectively bit length sum total, quantitative calibration sum total and number of pictures about all pictures before the 33rd picture in the 2nd subsequence.The picture rate is meant the still frame lattice number that per second is shown.

The second video signal complexity X _LDetermine via following formula two:

Formula two: $X_L=S_L*Q_L*\frac{F}{N_L^2}$

In the formula two, S _LRepresent one second bit length sum total, second bit length sum total is about in i the subsequence, the 1st bit length sum total to (j-1) individual picture; Q _LRepresent one second quantitative calibration sum total, second quantitative calibration sum total is about in i the subsequence, the 1st quantitative calibration sum total to (j-1) individual picture; N _LRepresent one second number of pictures, second number of pictures is about in i the subsequence, the 1st number of pictures to (j-1) individual picture.In the above embodiments, S _L, Q _LAnd N _LBe respectively about in the 2nd subsequence, the bit length of the 1st to the 32nd picture sum total, quantitative calibration sum total and number of pictures.

The initial quantization scale is determined by a predictive quantization scale (Predicted quantization scale) and a difference quantitative calibration (Differential quantization scale).Predictive quantization scale Q _PDetermine via following formula three:

Formula three: $Q_P=\frac{\mathrm{MIN}(X_A,X_L)}{B}+1$

In the formula three, X _ARepresent the first video signal complexity, the first video signal complexity is the video signal complexity about all pictures before j the picture in i the subsequence; X _LRepresent the second video signal complexity, the second video signal complexity is about in i the subsequence, the 1st video signal complexity to (j-1) individual picture; B represents a predetermined target bitrate.Aforesaid embodiment, X _AIt is video signal complexity about all pictures before the 33rd picture in the 2nd subsequence; X _LBe about in the 2nd subsequence, the video signal complexity of the 1st to the 32nd picture.

Difference quantitative calibration Q _AVia following formula four, determine:

Formula four: $Q_d=K*(\frac{S_{\mathrm{buf}}}{\mathrm{buf}\_\mathrm{size}}-0.5);$ Wherein $S_{\mathrm{buf}}=\mathrm{MAX}(S_{\mathrm{buf}}+S_j-\frac{B}{F},0)$

In the above-mentioned formula, S _BufRepresent bit stream buffer 18 present figure places of storing; S _jRepresent j the figure place that picture produced in present i the subsequence; Buf_size represents a predetermined bit stream buffer size (Buffer size); K represents one first pre-determined model parameter (Modelparameter), in order to determine the scale of this difference quantitative calibration.

Therefore, initial quantization scale Q can decide via following formula five:

Formula five: Q=MAX (Q _MIN, MIN (Q _MAX, Q _P+ Q _d)).

In the formula five, Q _MAXRepresent a predetermined maximum of this initial quantization scale, Q _MINRepresent a predetermined minimum value of this initial quantization scale.

See also Fig. 2 and Fig. 3, Fig. 2 is the calcspar of scene change detector 17 shown in Figure 1.The surveyed area schematic diagram that Fig. 3 is determined for decision module 171 shown in Figure 2.Scene change detector 17 of the present invention comprises a decision module (Determination module) 171 and one judge module (Judgment module) 172.Decision module 171 is used in present j the picture, determines a surveyed area (Detection area).Judge module 172 is used to surveyed area one last macro zone block place of row at present, judges that whether the present row in the surveyed area and in-line coding macro zone block (Intra macroblock) quantity in all first prostatitis sum up greater than a critical value (Threshold).If present row in the surveyed area and the in-line coding macro zone block quantity in all first prostatitis sum total are greater than critical value, then judge module 172 can judge that present j picture is a scene change.Otherwise if the in-line coding macro zone block quantity in the present row in the surveyed area and all first prostatitis is summed up not greater than critical value, then judge module 172 can continue present j picture judged, all row in surveyed area all detect and finish.Critical value THR_SC is determined by following formula six:

Formula six: $\mathrm{THR}\_\mathrm{SC}=\frac{N_{\mathrm{mbv}}}{\mathrm{DA}}*N_{\mathrm{mbh}}*\mathrm{SC}\_\mathrm{RATIO}+1$

In the formula six, N _MbvRepresent in present j the picture quantity of macro zone block that each row comprises; N _MbhRepresent in present j the picture quantity of macro zone block that each row comprises; DA is a natural number, and in order to the decision surveyed area, SC_RATIO is that a scene changes ratio (Scene change ratio).Aforesaid embodiment, in pre-defined 9 row of each picture of video signal sequence 12 and 11 row, that is each row all comprises 11 macro zone blocks, and each row all comprises 9 macro zone blocks.The size of big or small viewable pictures rate (F) in order to the DA of decision surveyed area adjusts.When DA is made as 3, and SC_RATIO was made as 20% o'clock, and critical value THR_SC is 7.6.

As shown in Figure 3, when DA was made as 3, surveyed area was 1/3 of whole image size.In this embodiment, the surveyed area of the 33rd picture in present the 2nd subsequence comprises three row r1, r2 and r3.Judge module 172 can be prior to last macro zone block place (the grey block among Fig. 3) of row r1, whether the in-line coding macro zone block quantity sum total of judging row r1 is greater than critical value, if, judge that then the 33rd picture in present the 2nd subsequence is a scene change, if not, continue then to judge that whether the in-line coding macro zone block quantity sum total of row r2 and row r1 is greater than critical value.All row in surveyed area all detect and finish, and continue the judgement of next picture again.

When scene change detector 17 detects j picture in i the subsequence and is a scene change, quantitative calibration generator 24 promptly can produce a quantitative calibration of adjusting based on the initial quantization scale, and encoder 16 can be encoded with the in-line coding pattern to j picture in i the subsequence again according to adjusted quantitative calibration.

Adjusted quantitative calibration

Determined by following formula seven:

Formula seven: $\hat{Q}=Q*(\frac{S_{\mathrm{int\; ra}}}{L}*\frac{N_{\mathrm{mb}}}{N_{\mathrm{int\; ra}}}*\frac{F}{B})$

In the formula seven, N _MbRepresent in present j the picture quantity of all macro zone blocks; N _IntraRepresent present j picture in surveyed area, the quantity of in-line coding macro zone block; S _IntraRepresent present j picture in surveyed area, the figure place of in-line coding macro zone block; L represents one second pre-determined model parameter.

The present invention also provides a kind of video encoding method, in order to a video signal sequence is encoded.The video signal sequence comprises N subsequence, and N is a natural number, and each subsequence all comprises a plurality of pictures.That is video encoding method of the present invention is in order to in the video signal sequence, and j picture in i subsequence encoded.Picture before this j picture in this i subsequence has all been finished coding, and i is the integer index in 1 to N scope, and j is between 2 integer indexs to the number of pictures scope of this i subsequence.Be noted that first picture in each subsequence is encoded with the in-line coding pattern.

See also Fig. 4, Fig. 4 is the flow chart of video encoding method of the present invention.Video encoding method of the present invention comprises the following step:

Step S100: beginning;

Step S102:, produce an initial quantization scale based on the described picture of having finished coding;

Step S104:, j picture in i the subsequence encoded with an intermediate code pattern according to the initial quantization scale;

Step S106: in present j picture, determine a surveyed area;

Step S108:, judge that whether the present row in the surveyed area and the in-line coding macro zone block quantity in all first prostatitis sum up greater than a critical value in surveyed area one last macro zone block place of row at present;

Step S110: if the result among the step S108 is certainly, judges that then present j picture is a scene change, and carry out step S112, if the result among the step S108 then carries out step S114 for negating;

Step S112: produce a quantitative calibration of adjusting based on the initial quantization scale, and, j picture in i the subsequence encoded again with an in-line coding pattern, then carry out step S116 according to adjusted quantitative calibration;

Step S114: continue present j picture carried out the judgement of step S108, all row in surveyed area all detect and finish, and carry out step S116;

Step S116: repeat step S102, till each picture is all finished coding in the video signal sequence;

Step S118: finish.

Compared to the file technology, in video coding system of the present invention and method thereof, when the present picture of one in the video signal sequence was a scene change, meeting of the present invention was carried out recompile according to the complexity information of before having finished coded picture to this present picture.In other words, the present invention only can carry out the operation of recompile at the picture of occurrence scene variation.Therefore, the present invention combines the technology that repeatedly changes bit rate and single change bit rate, makes video sequence be listed in through behind the coding, can obtain preferable, the consistent quality of image, and not need a large amount of memory spaces to come the memory encoding data.

By the above detailed description of preferred embodiments, be to wish to know more to describe feature of the present invention and spirit, and be not to come category of the present invention is limited with above-mentioned disclosed preferred embodiment.On the contrary, its objective is that hope can contain in the category of claim of being arranged in of various changes and tool equality institute of the present invention desire application.

Claims (24)

1. video encoding method, one video signal sequence comprises N subsequence, each subsequence all comprises a plurality of pictures, this video encoding method is in order to encode to j picture in i the subsequence, picture before this j the picture in this i subsequence, in this video signal sequence has all been finished coding, and N is a natural number, and i is the integer index in 1 to the N scope, j is 2 integer indexs to the number of pictures scope of this i subsequence, and this method comprises the following step:

(a), produce an initial quantization scale based on the described picture of having finished coding;

(b), this j the picture in this i subsequence encoded with one first coding mode according to this initial quantization scale;

(c) judge whether this j picture in this i subsequence (j-1) individual picture in this i subsequence on scene is a scene change; And

(d) if the result of step (c) for certainly, then produces one based on this initial quantization scale and adjusts quantitative calibration, and adjust quantitative calibration according to this, this j the picture in this i subsequence encoded again with one second coding mode.

2. the method for claim 1, wherein this first coding mode is an intermediate code pattern.

3. the method for claim 1, wherein this second coding mode is an in-line coding pattern.

4. the method for claim 1, wherein this initial quantization scale is determined by a predictive quantization scale and a difference quantitative calibration.

5. method as claimed in claim 4, wherein this predictive quantization scale Q _PDetermine via following formula:

$Q_P=\frac{\mathrm{MIN}(X_A,X_L)}{B}+1,$

X _ARepresent one first video signal complexity, this first video signal complexity is the video signal complexity about all pictures before this j the picture in this i subsequence, in this video signal sequence, X _LRepresent one second video signal complexity, this second video signal complexity is about in this i subsequence, the 1st video signal complexity to (j-1) individual picture, and B represents a predetermined target bitrate.

6. method as claimed in claim 5, wherein this first video signal complexity X _ADetermine via following formula:

$X_A=S_A*Q_A*\frac{F}{N_A^2},$

S _ARepresent one first bit length sum total, this first bit length sum total is to sum up Q about the bit length of all pictures before, in this video signal sequence of this j the picture in this i subsequence _ARepresent one first quantitative calibration sum total, this first quantitative calibration sum total is to sum up N about the quantitative calibration of all pictures before, in this video signal sequence of this j the picture in this i subsequence _ARepresent one first number of pictures, this first number of pictures is the number of pictures about all pictures before this j the picture in this i subsequence, in this video signal sequence; F represents a predetermined picture rate.

7. method as claimed in claim 6, wherein this second video signal complexity X _LDetermine via following formula:

$X_L=S_L*Q_L*\frac{F}{N_L^2},$

S _LRepresent a younger brother two bit lengths sum total, this second bit length sum total is about in this i subsequence, the 1st bit length sum total to (j-1) individual picture, Q _LRepresent one second quantitative calibration sum total, this second quantitative calibration sum total is about in this i subsequence, the 1st quantitative calibration sum total to (j-1) individual picture, N _LRepresent one second number of pictures, this second number of pictures is about in this i subsequence, the 1st number of pictures to (j-1) individual picture.

8. method as claimed in claim 7, wherein this difference quantitative calibration Q _dDetermine via following formula:

$Q_d=K*(\frac{S_{\mathrm{buf}}}{\mathrm{buf}\_\mathrm{size}}-0.5);$

Wherein $S_{\mathrm{buf}}=\mathrm{MAX}(S_{\mathrm{buf}}+S_j-\frac{B}{F},0),$

S _BufRepresent the present figure place of storing of a bit stream buffer, S _jRepresentative is this j figure place that picture produced at present, and buf_size represents a predetermined bit stream buffer size, and K represents one first pre-determined model parameter, in order to determine the scale of this difference quantitative calibration.

9. method as claimed in claim 8, wherein this initial quantization scale Q determines via following formula:

Q＝MAX(Q _MIN，MIN(Q _MAX，Q _P+Q _d))，

Q _MAXRepresent a predetermined maximum of this initial quantization scale, Q _MINRepresent a predetermined minimum value of this initial quantization scale.

10. the method for claim 1, wherein in each in these a plurality of pictures, the row of pre-defined some and row, each row comprises a plurality of macro zone blocks, and step (c) further comprises the following step:

(c1) in present this j picture, determine a surveyed area;

(c2), judge that being somebody's turn to do in this surveyed area is listed as at present and whether the in-line coding macro zone block quantity in all first prostatitis sums up greater than a critical value in last macro zone block place of these surveyed area one present row; And

(c3) if the result of step (c2) is for affirming, judge that then this j picture is a scene change at present, if the result in the step (c2) then continues present this j picture carried out the judgement of step (c2) for negating, all row in surveyed area all detect and finish.

11. method as claimed in claim 10, wherein this critical value is determined by following formula:

$\mathrm{THR}\_\mathrm{SC}=\frac{N_{\mathrm{mbv}}}{\mathrm{DA}}*N_{\mathrm{mbh}}*\mathrm{SC}\_\mathrm{RATIO}+1$

Wherein, THR_SC represents this critical value, N _MbvIn present this j the picture of representative, the quantity of macro zone block that each row comprises, N _MbhIn present this j the picture of representative, the quantity of macro zone block that each row comprises, DA is a natural number, in order to determine this surveyed area, SC_RATIO is that a scene changes ratio.

12. method as claimed in claim 11 wherein should be adjusted quantitative calibration

Determined by following formula:

$\hat{Q}=Q*(\frac{S_{\mathrm{int\; ra}}}{L}*\frac{N_{\mathrm{mb}}}{N_{\mathrm{int\; ra}}}*\frac{F}{B}),$

N _MbIn present this j the picture of representative, the quantity of all macro zone blocks, N _IntraDuring present this j the picture of representative encoded the first time, the quantity of in-line coding macro zone block, S _IntraRepresent present j picture in surveyed area, the figure place of in-line coding macro zone block, L represents one second pre-determined model parameter.

13. video coding system, one video signal sequence comprises N subsequence, each subsequence all comprises a plurality of pictures, this video coding system is in order to encode to j picture in i the subsequence, picture before this j the picture in this i subsequence, in this video signal sequence has all been finished coding, and N is a natural number, and i is the integer index in 1 to the N scope, j is 2 integer pointers to the number of pictures scope of this i subsequence, and this system comprises:

One quantitative calibration generator, this quantitative calibration generator are based on the described picture of having finished coding, in order to produce an initial quantization scale;

One encoder, this encoder are coupled in this quantitative calibration generator, and according to this initial quantization scale, in order to this j the picture in this i subsequence encoded with one first coding mode; And

One scene change detector, this scene change detector is coupled in this encoder, and in order to judge whether this j picture in this i subsequence (j-1) individual picture in this i subsequence on scene is a scene change;

Wherein, if this j the picture in this i subsequence is a scene change, then this quantitative calibration generator produces one based on this initial quantization scale and adjusts quantitative calibration, and this encoder is adjusted quantitative calibration according to this, and this j the picture in this i subsequence encoded again with one second coding mode.

14. system as claimed in claim 13, wherein this first coding mode is an intermediate code pattern.

15. system as claimed in claim 13, wherein this second coding mode is an in-line coding pattern.

16. system as claimed in claim 13, wherein this initial quantization scale is determined by a predictive quantization scale and a difference quantitative calibration.

17. system as claimed in claim 16, wherein this predictive quantization scale Q _PDetermine via following formula:

$Q_P=\frac{\mathrm{MIN}(X_A,X_L)}{B}+1,$

X _ARepresent one first video signal complexity, this first video signal complexity is the video signal complexity about all pictures before this j the picture in this i subsequence, in this video signal sequence; X _LRepresent one second video signal complexity, this second video signal complexity is about in this i subsequence, the 1st video signal complexity to (j-1) individual picture; B represents a predetermined target bitrate.

18. system as claimed in claim 17, wherein this first video signal complexity X _ADetermine via following formula:

$X_A=S_A*Q_A*\frac{F}{N_A^2},$

S _ARepresent one first bit length sum total, this first bit length sum total is to sum up about the bit length of all pictures before, in this video signal sequence of this j the picture in this i subsequence; Q _ARepresent one first quantitative calibration sum total, this first quantitative calibration sum total is to sum up about the quantitative calibration of all pictures before, in this video signal sequence of this j the picture in this i subsequence; N _ARepresent one first number of pictures, this first number of pictures is the number of pictures about all pictures before this j the picture in this i subsequence, in this video signal sequence; F represents a predetermined picture rate.

19. system as claimed in claim 18, wherein this second video signal complexity X _LDetermine via following formula:

$X_L=S_L*Q_L*\frac{F}{N_L^2},$

S _LRepresent one second bit length sum total, this second bit length sum total is about in this i subsequence, the 1st bit length sum total to (j-1) individual picture; Q _LRepresent one second quantitative calibration sum total, this second quantitative calibration sum total is about in this i subsequence, the 1st quantitative calibration sum total to (j-1) individual picture; N _LRepresent one second number of pictures, this second number of pictures is about in this i subsequence, the 1st number of pictures to (j-1) individual picture.

20. system as claimed in claim 19, wherein this difference quantitative calibration Q _dDetermine via following formula:

$Q_d=K*(\frac{S_{\mathrm{buf}}}{\mathrm{buf}\_\mathrm{size}}-0.5);$

Wherein $S_{\mathrm{buf}}=\mathrm{MAX}(S_{\mathrm{buf}}+S_j-\frac{B}{F},0),$

S _BufRepresent the present figure place of storing of a bit stream buffer, S _jRepresentative is this j figure place that picture produced at present, and buf_size represents a predetermined bit stream buffer size, and K represents one first pre-determined model parameter, in order to determine the scale of this difference quantitative calibration.

21. method as claimed in claim 20, wherein this initial quantization scale Q determines via following formula:

Q＝MAX(Q _MIN，MIN(Q _MAX，Q _P+Q _d))，

Q _MAXRepresent a predetermined maximum of this initial quantization scale, Q _MINRepresent a predetermined minimum value of this initial quantization scale.

22. system as claimed in claim 13, wherein in each in these a plurality of pictures, the row of pre-defined some and row, each row comprises a plurality of macro zone blocks, and this scene change detector further comprises:

One decision module, this decision module are used to determine a surveyed area in present this j picture; And

One judge module, this judge module are used to last macro zone block place of these surveyed area one present row, judge that being somebody's turn to do in this surveyed area is listed as at present and whether the in-line coding macro zone block quantity in all first prostatitis sums up greater than a critical value;

Wherein, if the in-line coding macro zone block quantity sum total in the row at present in this surveyed area and all first prostatitis is greater than this critical value, then this judge module judges that this j picture is a scene change at present, otherwise, then this judge module continues present this j picture judged, all row in surveyed area all detect and finish.

23. the system as claimed in claim 22, wherein this critical value is determined by following formula:

$\mathrm{THR}\_\mathrm{SC}=\frac{N_{\mathrm{mbv}}}{\mathrm{DA}}*N_{\mathrm{mbh}}*\mathrm{SC}\_\mathrm{RATIO}+1,$

Wherein, THR_SC represents this critical value, N _MbvIn present this j the picture of representative, the quantity of macro zone block that each row comprises, N _MbhIn present this j the picture of representative, the quantity of macro zone block that each row comprises, DA is a natural number, in order to determine this surveyed area, SC_RATIO is that a scene changes ratio.

24. system as claimed in claim 23 wherein should adjust quantitative calibration

Determined by following formula:

$\hat{Q}=Q*(\frac{S_{\mathrm{int\; ra}}}{L}*\frac{N_{\mathrm{mb}}}{N_{\mathrm{int\; ra}}}*\frac{F}{B}),$

N _MbIn present this j the picture of representative, the quantity of all macro zone blocks, N _IntraDuring present this j the picture of representative encoded the first time, the quantity of in-line coding macro zone block, S _IntraRepresent present j picture in surveyed area, the figure place of in-line coding macro zone block, L represents one second pre-determined model parameter.

Images